OpenAI has announced Aardvark, a new AI agent that can help developers and security teams detect and fix vulnerabilities at scale.
Aardvark, driven by GPT-5, is said to monitor commits, build threat models, analyze the behavior of code, and perform tests inside sandboxed environments for confirmation of real-world exploitability.
The system integrates with GitHub and OpenAI Codex to offer annotated explanations, human-review patches, and low false-positive insights that enhance defensive posture without slowing the velocity of development workflows.
Aardvark detected 92% of the known and synthetic vulnerabilities in testing, finding complex issues that only arise under rare conditions.
It has already discovered multiple bugs in open-source projects, ten of which have received CVE identifiers.
OpenAI promises to provide pro-bono scanning to select non-commercial repositories and updated its disclosure policy to emphasize collaboration over pressure.
As software vulnerabilities continue to scale, Aardvark represents a defender-first approach, giving organizations continuous protection as code evolves.
You may also want to check out some of our other recent updates.
Wanna know what’s trending online every day? Subscribe to Vavoza Insider to access the latest business and marketing insights, news, and trends daily with unmatched speed and conciseness! 🗞️
